christoph/fail
1
Fork 0
Code Actions Packages Releases Activity

Commit Graph

  • fa1875e2f2 #cmakedefine does not make sense for non-binary configuration options hsc 2012-11-20 15:01:46 +00:00
  • 35026de2d8 ecos: use multiple intermediate states to speed up experiments (disabled) hsc 2012-11-20 15:01:43 +00:00
  • 041746741d ecos: include variant and benchmark in job info hsc 2012-11-20 15:01:40 +00:00
  • 20fc14e6b3 ecos: result merging repaired hsc 2012-11-20 15:01:37 +00:00
  • a1e99f9004 Pointers back in - this solution seems to work for now unzner 2012-11-20 13:25:59 +00:00
  • 6d340113cd The pointer structure caused weird server errors, so I will only maintain it for the methods to avoid unnecessary parameters. unzner 2012-11-20 11:11:43 +00:00
  • 4db7272e5f Two minor bugfixes unzner 2012-11-20 09:58:44 +00:00
  • 49d1608969 correct sanity checks for client/server communication hsc 2012-11-14 13:31:53 +00:00
  • b8a4797360 doc: REP/REPZ/REPNZ done right hsc 2012-11-13 16:27:07 +00:00
  • 4f48eb3232 ecos: send mandatory ecos_test_result when sanity check fails hsc 2012-11-13 16:26:58 +00:00
  • 4cce602bca ecos: actually enqueue jobs hsc 2012-11-13 00:17:33 +00:00
  • 7b3e5986d1 ecos: save experiment runtime hsc 2012-11-13 00:17:31 +00:00
  • 077f7ea507 ecos: load existing results on campaign startup hsc 2012-11-13 00:17:28 +00:00
  • 61ab977f01 bugfix: read(2) returns 0 on EOF hsc 2012-11-13 00:17:25 +00:00
  • 97299ba370 util/WallclockTimer: (float) cast operator hsc 2012-11-13 00:17:22 +00:00
  • 9decf95a09 ecos: specific traps are OK for two benchmarks hsc 2012-11-13 00:17:18 +00:00
  • a0e6ddd519 ecos: fault-space cutoff specific for this campaign hsc 2012-11-12 13:55:45 +00:00
  • 6f98d64613 bugfix: racecondition removed hellwig 2012-11-12 11:46:26 +00:00
  • 464bc9390b ecos: campaign rewritten hsc 2012-11-12 10:59:35 +00:00
  • a02088b33e enable support for large (>2G) files hsc 2012-11-12 10:59:32 +00:00
  • 283d946295 ecos: no need for experimentInfo.hpp.sh anymore hsc 2012-11-12 10:59:28 +00:00
  • 1b5bbba0e5 Handled a kind of wicked error case which nevertheless happens quite frequently unzner 2012-11-11 11:37:08 +00:00
  • 35b1d0203e CampaignManager: destructor / cleanup hsc 2012-11-10 16:18:40 +00:00
  • 86ba9cb377 CampaignManager: only instantiate JobServer when needed hsc 2012-11-10 16:14:06 +00:00
  • 4957cfb71b ecos: remove hard-coded trace filename hsc 2012-11-10 16:13:59 +00:00
  • 0ab4f12b7c ecos: no need for clearListeners anymore hsc 2012-11-09 15:37:33 +00:00
  • 661563125e ecos: cosmetics / compiler warning hsc 2012-11-09 15:37:27 +00:00
  • 1292b66043 ecos: image/prerequisite/result path naming changes hsc 2012-11-09 15:37:24 +00:00
  • c2242f2bb4 ecos: guest-image independent fail client hsc 2012-11-09 15:37:21 +00:00
  • 805f67e23c ElfReader: return ADDR_INV instead of (signed) -1 on failure hsc 2012-11-09 15:37:18 +00:00
  • a26aeb52ee ecos: preparations for guest-image independent fail client hsc 2012-11-09 15:37:14 +00:00
  • bbf7731aa2 fast-watchpoints: implementation finished (and tested). adrian 2012-11-08 13:36:08 +00:00
  • ddbab2e903 build-system: separated fast-breakpoints, added fast-watchpoints. adrian 2012-11-08 13:35:22 +00:00
  • 4a6b576b18 fast-breakpoints: created template class DefPerfVector, realizing the buffer interface. Modified PerfVectorBreakpoints appropriately. adrian 2012-11-08 13:18:40 +00:00
  • 148b0bb483 fast-breakpoints: work even if fast-watchpoints are enabled (TODO removed). adrian 2012-11-08 13:07:02 +00:00
  • 5648093ffe fast-breakpoints: use explicit namespace qualifier in aspect header. adrian 2012-11-08 13:05:05 +00:00
  • e337fef205 perf-test: further results added adrian 2012-11-08 12:59:43 +00:00
  • 2eb18c81ac perf-test: code improved. adrian 2012-11-08 12:59:16 +00:00
  • 126fb1a1ff ecos_kernel_test: remove 2nd reboot (which was just before a restore) chb 2012-11-07 16:01:35 +00:00
  • 4cef94a016 Removed the recommendation of repeat-speedups to avoid misunderstandings (see revision 1885) unzner 2012-11-07 09:02:12 +00:00
  • 9c3755363b bochs bugfix: store the timer ID for correct deregistration hsc 2012-11-06 18:29:45 +00:00
  • 1668e21ce1 bochs: removed preprocessor nonsense hsc 2012-11-06 18:29:41 +00:00
  • 1f8dfd01e9 bochs: trace REP/REPZ/REPNZ iterations hsc 2012-11-06 15:43:24 +00:00
  • 0b8710872b Removed BochController debug stuff. Merged: BochsController::onBreakpoint -> SimCon::onBreakpoint. adrian 2012-11-06 11:40:05 +00:00
  • b7bbcd389f ecos_kernel_test: record error detected/corrected chb 2012-11-06 10:51:41 +00:00
  • 9e5de7b14a ecos_kernel_test bugfix: timer granularity of 10s is way too coarse hsc 2012-11-02 16:59:07 +00:00
  • cb3775b290 Ensuring parsability in case of exceptions by removing unnecessary newline characters unzner 2012-11-01 13:28:52 +00:00
  • 1423fb93dd Removed instruction vector to reduce memory consumption unzner 2012-11-01 09:23:45 +00:00
  • 163be81874 rampage: final experiment parameters, for the records hsc 2012-10-31 15:07:01 +00:00
  • e1dcbece88 limit fail-client instances on "big" IRB servers by default hsc 2012-10-31 15:06:58 +00:00
  • 4ec51b003a bugfix: WallclockTimer::operator<< must be inlined to prevent multiple defs. adrian 2012-10-31 12:42:48 +00:00
  • dd17034fce WallclockTimer: coding style unifications. adrian 2012-10-30 12:15:26 +00:00
  • ccc4aca728 Forgot to cast the type unzner 2012-10-30 11:09:59 +00:00
  • 22634f6103 Minor bugfix unzner 2012-10-30 10:59:27 +00:00
  • 7ae8b6b09d Introduces instruction filtering, e.g. to filter for kernel instructions unzner 2012-10-30 09:14:33 +00:00
  • 247fed5aa7 Gem5: Implemented MemoryManager friemel 2012-10-28 23:50:08 +00:00
  • 66fe662495 Gem5: functionalAccess() should not trigger memory access friemel 2012-10-28 23:50:06 +00:00
  • 4a48b6c20c nanojpeg: deleting result sets after handling them is not racy hsc 2012-10-27 23:18:27 +00:00
  • 4e26205aef nanojpeg: count existing experiments and CSV rows hsc 2012-10-26 17:31:44 +00:00
  • 8f39aa9717 nanojpeg: correct experiment count hsc 2012-10-26 17:31:41 +00:00
  • 9038e7f603 increase pick window size for job reassignments hsc 2012-10-26 17:31:38 +00:00
  • 720a59cfde runcampaign.sh uses start-clients.sh instead of duplicating code hsc 2012-10-26 17:31:35 +00:00
  • 3353bad6c6 number of clients per CPU = 1 hsc 2012-10-26 17:31:31 +00:00
  • 3dcc684400 read(2)/write(2) wrappers for reliable delivery hsc 2012-10-26 16:13:41 +00:00
  • 55dd79cc03 cosmetics hsc 2012-10-26 16:13:36 +00:00
  • 5a5367beee nanojpeg: complete fault-space coverage by closing ECs in the end hsc 2012-10-26 14:27:35 +00:00
  • 05217d77d9 nanojpeg: more thorough output address check hsc 2012-10-26 14:27:32 +00:00
  • 34d89000cf ecos_kernel_test: simplification (no need to call clearListeners() here). adrian 2012-10-25 15:52:55 +00:00
  • e8f715f8af comment-fix (not comprehensible this way) adrian 2012-10-25 15:51:20 +00:00
  • 6b1f457662 bugfix: onDeletion() cannot be called on listeners in the fire-list. (+ simplified makeActive().) adrian 2012-10-25 15:49:25 +00:00
  • 12f3550428 Reboot-test optimized hellwig 2012-10-25 12:39:09 +00:00
  • 1eb8ac6870 Bugfix: missing header added hellwig 2012-10-25 12:36:56 +00:00
  • 22cd362cc7 gem5: only enable Gem5Listener.ah when building gem5 hsc 2012-10-24 23:40:10 +00:00
  • 14ca377a85 nanojpeg: run 200 experiments, not 50 jobs hsc 2012-10-24 22:35:49 +00:00
  • bd3bef1f72 nanojpeg: load existing results before starting campaign hsc 2012-10-24 22:35:46 +00:00
  • 4fb768972b nanojpeg: allow CSV output to be buffered hsc 2012-10-24 22:35:43 +00:00
  • 6c204d2a93 don't distribute to lidong by default hsc 2012-10-24 22:32:05 +00:00
  • e0e95faa5b Restructured the gem5 backend: - FailGem5Device is gone. - There are now changes directly made to the gem5 source. - Gem5Connector is a helper class that is compiled inside the gem5 context to workaround problems with gem5 header in fail. friemel 2012-10-24 19:19:14 +00:00
  • b41eec3f65 Adding gem5 source to svn. friemel 2012-10-24 19:18:57 +00:00
  • f7ff71bd46 Revert "ListenerManager bugfix (segfault): [...]" hsc 2012-10-24 15:40:57 +00:00
  • 8588617a86 ListenerManager bugfix (segfault): clear fire-list when clearListener() is called (this revealed a bug in reboot() which needs to be fixed too). adrian 2012-10-24 13:42:05 +00:00
  • 4dcee177be nanojpeg: receive/store results, merge identical neighbors hsc 2012-10-24 09:58:59 +00:00
  • 554d80139a nanojpeg: timeout reduced, instructions limited (temporarily) hsc 2012-10-24 09:58:56 +00:00
  • f61db1ff39 nanojpeg: psnr is optional hsc 2012-10-24 09:58:53 +00:00
  • 1585b4a614 nanojpeg: signedness fixes hsc 2012-10-24 09:58:50 +00:00
  • 8808eabc3c bugfix: multiple CMAKE_AGPP_FLAGS hsc 2012-10-24 09:58:48 +00:00
  • bb621aac82 nanojpeg: uint32_t for IPs in protobuf message hsc 2012-10-24 09:58:45 +00:00
  • 3de22e1b10 nanojpeg: count both jobs and single experiments hsc 2012-10-24 09:58:42 +00:00
  • 2b1e4edd6b nanojpeg: store known-outcome results, enqueue jobs hsc 2012-10-24 09:58:39 +00:00
  • 892e5f86f0 nanojpeg: use absolute IP from trace hsc 2012-10-24 09:58:36 +00:00
  • b167b21de6 nanojpeg: handle known-outcome ECs hsc 2012-10-24 09:58:33 +00:00
  • 6af5689cce nanojpeg: cosmetics, updated CSV header hsc 2012-10-24 09:58:26 +00:00
  • f74954f460 nanojpeg: more diagnostic info for weird BROKEN cases hsc 2012-10-23 13:36:06 +00:00
  • c5304c0c31 nanojpeg: detect rogue memory accesses hsc 2012-10-23 13:36:03 +00:00
  • 8de290f47d nanojpeg: experiment largely complete hsc 2012-10-22 21:13:20 +00:00
  • 781fbf1f4e elfreader: be less verbose hsc 2012-10-22 21:13:16 +00:00
  • 41cf5e1352 nanojpeg: campaign's fault-space pruning works now hsc 2012-10-22 15:36:18 +00:00
  • 1ae19893ea nanojpeg: debug output (missing opcode information) hsc 2012-10-22 15:36:15 +00:00
  • d26fc600da nanojpeg: helper for Udis86 -> Fail* register ID translation hsc 2012-10-22 15:36:12 +00:00
  • 0d1618746c sal/bochs: move register IDs to a separate file hsc 2012-10-22 15:36:09 +00:00